Diversity Measures and Coarse-graining in Data Analysis with an Application Involving Plant Species on the Gal´apagos Islands
Radu Cornel Guiasu, Silviu Guiasu
In a numerical entity-characteristic incidence ma-
trix we can use simple or multiple regression
and calculate correlations between pairs of char-
acteristics. However, in order to detect similari-
ties/dissimilarities, interdependence, and multiple
probabilistic causality among the characteristics we
have to group the entities in classes. The num-
ber of uniform classes obtained by coding the given
values of these characteristics depends on the bal-
ance between the class uncertainty and class ambi-
guity. The similarity, interdependence, and multi-
ple probabilistic causality among characteristics are
analyzed. When a set of entities and the abundance
of their components are given, the average within-
entity diversity and the average between-entity di-
versity are studied. The results are applied to the
number of endemic and immigrant plant species in
the Gal´apagos Islands. Full Text
|